CODEWITHSUNDEEP
pythonpandasnumpy
Álvaro
Álvaro

Reputation: 2588

Groupby then check row match and count concurrent instances for that value

I have this DataFrame

     car   color  years  max_years
0   audi   black      1          7
1   audi    blue      2          7
2   audi  purple      4          7
3   audi   black      6          7
4    bmw    blue      1          5
5    bmw    green     2          5
6    bmw    grey      5          5
7    bmw    blue     20          5
8   fiat   green      1          4
9   fiat   green      3          4
10  fiat   green      4          4
11  fiat   green     10          4

If a color entry is 1 year I want to count how many more times that color appears for that car brand group up to the max years for that group.

I would like to run the isin color condition for each car brand group, I think my problem is that the color list is not grouby('car') and therefore the evaluation is for all cars

The result should be:

0  audi       2
1   bmw       1
2  fiat       3

Any help would be appreciated


import pandas as pd

car =  ['audi', 'audi', 'audi', 'audi', 'bmw', 'bmw', 'bmw', 'bmw', 'fiat', 'fiat', 'fiat', 'fiat']
color =  ['black', 'blue', 'purple', 'black', 'blue', 'green', 'grey', 'blue', 'green', 'green', 'green', 'green']
years = [1, 2, 4, 6, 1, 2, 5, 20, 1, 3, 4, 10, ]
max_years = [7, 7, 7, 7, 5, 5, 5, 5, 4, 4, 4, 4]

data = {'car': car, 'color': color, 'years': years, 'max_years': max_years}
df = pd.DataFrame(data=data)

colors =  df.loc[df.years == 1]['color'].values

colour_cars = df[(df.years <= df.max_years) & df['color'].isin(colors)].groupby(['car']).size().reset_index(name='colour_cars')

print(colour_cars)

Upvotes: 2

Views: 145

Answers (1)

jezrael
jezrael

Reputation: 862681

Idea is use Series.map by Series created with filtered DataFrame with years == 1 and compare by column color:

colors =  df.loc[df.years == 1].set_index('car')['color']

df1 = (df[(df.years <= df.max_years) & df['car'].map(colors).eq(df['color'])]
         .groupby('car')
         .size()
         .reset_index(name='colour_cars'))
print(df1)

    car  colour_cars
0  audi            2
1   bmw            1
2  fiat            3

Or you can use mask converted to integers by Series.view, then is necessary count Trues values by sum and pass Series df['car'] to groupby:

colors =  df.loc[df.years == 1].set_index('car')['color']

df1 = (((df.years <= df.max_years) & df['car'].map(colors).eq(df['color']))
         .view('i1')
         .groupby(df['car'])
         .sum()
         .reset_index(name='colour_cars'))
print(df1)

    car  colour_cars
0  audi            2
1   bmw            1
2  fiat            3

Different idea is test first color per group by GroupBy.transform with first (solution working if always first year per group is 1):

df2 = (df[(df.years <= df.max_years)]
           .groupby('car')['color']
           .transform('first').eq(df['color'])
           .view('i1')
           .groupby(df['car'])
           .sum()
           .reset_index(name='colour_cars'))

print(df2)

    car  colour_cars
0  audi            2
1   bmw            1
2  fiat            3

Upvotes: 4

Related Questions